[clang] Add tests for CWG issues 6, 212, 232, 2823. (PR #165633)
Vlad Serebrennikov via cfe-commits
cfe-commits at lists.llvm.org
Thu Oct 30 14:25:47 PDT 2025
================
@@ -230,6 +230,37 @@ namespace cwg211 { // cwg211: 2.7
};
} // namespace cwg211
+namespace cwg212 { // cwg212: yes
----------------
Endilll wrote:
> Has this changed to mean just "we don't know when we started supporting this" instead?
I've definitely seen this interpretations in PRs that added DR tests without fixing anything. I pushed them in the direction of actually figuring out at which point Clang started to work the way the wording suggests.
> I don't think that's an improvement: saying "2.7" here suggests that version 2.6 did something else, and a change was made in 2.7 to fix this, which isn't really true.
I don't disagree, but I've seen DRs like CWG182, which had `yes` status, but it turned out that anyone with Clang 13 or earlier didn't get the behavior described in the response. While `Clang 2.7` can give a wrong impression to people who don't know that this was the first version with C++ support that didn't require GCC, getting out of habit of slapping `yes` on tests made DR status page more useful to assess conformance.
> But if that's what we're doing now, I guess it makes sense to be consistent with that.
I'm taking full blame for the changes made in this area in the past couple of years, as I was the one pushing them. But specifying the version would indeed be more consistent with status quo. I see there are 3 `Yes` statuses on the page, which is my overlook.
https://github.com/llvm/llvm-project/pull/165633
More information about the cfe-commits
mailing list